If a length of $release and/or $versionsuffix is more than 8 bytes,
a length of DLL name may be more than 8.
Then, this corrupts a generated DLL on OS/2.
This patch assures 3 bytes for libname, 2 bytes for release, and 3 bytes
for version at least.
* m4/libtool.m4 (soname_spec) [os2*]: Limit a length of DLL name to 8.3
correctly.
old_archive_from_new_cmds is not used at all. In addition, if this is
defined static libs are not generated.
This is a regression of commit e60044.
* m4/libtool.m4: Remove old_archive_from_new_cmds on OS/2.
The check for Objective C++ was combined with the check for
Objective C, which caused Objective C++ tests to fail if language
support was enabled for OBJC but not OBJCXX. Additionally, a syntax
error with macros LT_PROG_OBJC and LT_PROG_OBJCXX was observed on
Fedora, exhibiting differing behaviour than observed on other
GNU/Linux operating systems.
Reported: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=79468
* m4/libtool.m4: Add separate check for Objective C++.
* tests/flags.at, tests/infer-tag.at: Update to check cached variable
for Objective C++, objcxx_compiles.
The Microsoft Macro Assembler can be used to assemble and link one
or more assembly-language source files with libtool.
Reported: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=48993
* Makefile.am: Pass ML64 flags to tests.
* configure.ac: Add language support for the Microsoft Macro Assembler.
* doc/libtool.texi: Update documentation to include ML64.
* m4/libtool.m4: Enable macro support for ML64.
* tests/flags.at: Add test for ML64 flags.
* tests/infer-tag.at: Add test for ML64 tag.
* NEWS: Update.
Reported: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=78817
* libltdl/libltdl/lt__alloc.h: Change FREE and MEMREASSIGN macros for
-Wcompound-token-split-by-macro.
* libltdl/libltdl/lt_system.h: Change LT_STMT_START and LT_STMT_END
macros for -Wcompound-token-split-by-macro.
* libltdl/loaders/loadlibrary.c: Add support for icx, icpx, ifx.
* libltdl/ltdl.c: Fixes for Clang.
* m4/libtool.m4: Add support for icx, icpx, and ifx.
* NEWS: Update.
Without cygpath, Window's 'cmd' may hang with a single slash (when transforming)
and may hang with two slashes (when not transforming).
Reported: https://savannah.gnu.org/support/index.php?110901
* build-aux/ltmain.in: Use one or two slashes with Window's command
prompt based on cached value in lt_cv_cmd_slashes.
* m4/libtool.m4: Cache whether one or two slashes are needed with
Window's command prompt, when cygpath is not installed in path.
lt_ar_flags are unquoted, which causes failures when additional options
are passed to ARFLAGS.
Reported: https://savannah.gnu.org/support/?111257
* m4/libtool.m4: Make lt_ar_flags declare as quoted.
MSYS does not have cygpath installed by default, so during configure,
a check for cygpath is cached. If cygpath is not found, Window's command
prompt will be used with a forward slash escape on the 'c' flag:
"cmd //c".
Reported: https://savannah.gnu.org/support/?110901
* build-aux/ltmain.in: Choose file path conversion for MSYS(2) by
whether cygpath is installed.
* m4/libtool.m4: Cache whether cygpath is installed in user path.
MSVC's compiler uses the .exp entension for their own export format,
which causes a build failure when a subsequent build is run. The
MSVC compiler, cl.exe, errors for an invalid file format. This is fixed
by using a similar extension name to avoid the naming collision.
Reported: https://savannah.gnu.org/support/?111157
* build-aux/ltmain.in: Replace .exp extension with .expsym.
* m4/libtool.m4: Replace .exp extension with .expsym.
* m4/libtool.m4 (global_symbol_to_import, global_symbol_to_cdecl,
global_symbol_to_c_name_address,
global_symbol_to_c_name_address_lib_prefix): List valid C symbol names
only in lt_<modname>_LTX_preloaded_symbols array.
On Cygwin, gawk-4.1.4-3 and newer do not treat the input stream as text
any more, leaving us behind with carriage return in symbol names.
* m4/libtool.m4 (lt_cv_sys_global_symbol_pipe): In AWK script for MS
dumpbin drop CR first.
With MSVC, linker flags were not properly being linked in during a
compile command. A check has been added to verify if there are linker
flags set and then use a different command when linking.
Fixes applied for bug#74175.
* build-aux/ltmain.in: Fix command parser for '-Wl' option for MSVC.
* m4/libtool.m4: Fix MSVC compile commands.
* tests/flags.at: Use '$GREP --' instead of '$FGREP' when comparing with
stdout.
Libtool checks only for libraries linked as -l* when trying to
find internal compiler libraries. Clang, however uses the absolute
path to link its internal libraries e.g. compiler_rt. This patch
handles clang's statically linked libraries when finding internal
compiler libraries.
https://crbug.com/749263https://debbugs.gnu.org/cgi/bugreport.cgi?bug=27866
Add configuration options to choose whether to use '-nostdlib' when
linking C++ shared libraries and modules.
* m4/libtool.m4: Replace '-nostdlib' with variable '$stdlibflag' and
set '$output_verbose_link_cmd' to be empty when enabling standard
library linking for C++ modules.
* m4/ltoptions.m4: Add new configuration options, --enable-cxx-stdlib
and --disable-cxx-stdlib, and add LT_INIT options, cxx-stdlib and
no-cxx-stdlib.
* doc/libtool.texi: Update for new configuration options.
* NEWS: Update.
commit:c7be1f229 addresses an issue with building C++ modules on NetBSD, by
removing -lgcc_s and -lgcc. This may sometimes not work correctly, depending
on the architecture, as libgcc_s.so and libgcc.a may contain required code to
properly link the shared object. On NetBSD, libgcc.a is compiled with -fPIC
and can be linked to a shared object without any issue.
Reported: https://savannah.gnu.org/support/index.php?111210
Omitting lines starting with 'Driving:' fixes the invalid postdeps
observed with FC.
Reported: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=76177
* m4/libtool.m4: Omit lines that start with 'Driving:' when setting
output_verbose_link_cmd.
* m4/libtool.m4: Linking a shared version of 'gcc' is not possible on
NetBSD, since the shared version of the library is 'gcc_s'. postdeps
have been updated to remove 'gcc' instances with sed.
The option --with-aix-soname=svr4 does not work the same as using
LT_INIT([aix-soname=svr4]), since the configure script is reordered
depending on which is used. The reordering causes 'enable_shared'
to not be set before processing for 'aix-soname' when using LT_INIT.
For LT_INIT([aix-soname=svr4]) to be set as intended, 'shared' must
be set before 'aix-soname=svr4', like LT_INIT([shared aix-soname=svr4]).
* m4/ltoptions.m4: Warn user if 'enable_shared' has not been set when
setting 'aix-soname' and add configuration messages to show which
variant of shared library versioning has been set for 'aix-soname'.
* doc/libtool.texi: Update to specify need for 'shared' to be used
before 'aix-soname' for LT_INIT.
Reported: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=67754
* Makefile.am: Set environment variables for tests.
* configure.ac: Add language support for Objective-C and Objective-C++.
* build-aux/ltmain.in: Update help message for tags available.
* m4/libtool.m4: Add tagging support for OBJC and OBJCXX.
* tests/flags.at: Add tests for passing OBJC and OBJCXX flags.
* tests/infer-tag.at: Add tests for inferring OBJC and OBJCXX tags.
* doc/libtool.texi: Update documentation.
* NEWS: Update.
The main Ironclad kernel-based distribution (Gloire) uses the mlibc
C library for its userland, and that is what commit
bd15931e57f5171732a1a911cbbad9b5e03d0308 was intended to support.
There is no reason this support should be Ironclad-specific, as there
are many other OSes that use mlibc, most prominently "Managarm",
which is already supported by gnu-config.
This patch adds support for emscripten compiler for shared compilation.
* build-aux/ltmain.in: clone link-mode handling for emscripten from
unixware7.
* m4/libtool.m4: translate emscripten specific flags for shared module
building.
This hobbyist OS has already been added to `config.sub` (and
`config.guess` respectively) some time ago, but was still lacking
upstream support for building libraries using libtool.
Since it is a relatively up-to-date system with ports of modern
software, "adding support" mostly just means adding empty cases to avoid
falling though to the most basic behavior (that guarantees compatibility
at the expense of disabling everything that might be critical).
* m4/libtool.m4: Add support for SerenityOS.
A space after '-Fe' treats a file name as an input file instead of
as the intended output file.
* m4/libtool.m4: Remove space after '-Fe'.
* NO-THANKS: Add Kirill Makurin for bug report 74175.
* NEWS: Update.
When generating an executable in OpenBSD, the executable fails to
generate when shlibpath_var, LD_LIBRARY_PATH, is used solely to
locate dependent libraries.
For hardcoding library paths in OpenBSD, see documentation for
hardcode_direct and hardcode_libdir_flag_spec.
* m4/libtool.m4: Update hardcode_shlibpath_var from 'no' to 'unsupported'
by removing 'with_gnu_ld=no' for OpenBSD.
The GCC/binutils tools support response files just fine, while
lld (impersonating GNU ld) only supports response files, not
linker scripts. Using a linker script as input just to pass a
list of files is overkill for cases when a response file is enough.
* build-aux/ltmain.in: Move creation of ld script.
* m4/libtool.m4: Add new variable tag for file_list_spec.
As said in the Autoconf source, the AC_TRY_EVAL macro is dangerous and
undocumented, and should not be used.
In particular, the one related to nm yields binary data in the config.log
file with dash, where "echo \\1" (echo with the argument \1) produces the
control character ^A instead of the usual \1 with most shells (POSIX says
that the result is implementation-defined). See:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=910076
This patch attempts to replace this AC_TRY_EVAL occurrence by code with
similar behavior, but using $ECHO instead of echo in order to avoid the
backslash issue.
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=21309
If $multilib is set to more than one word, the following error can be
seen during configuration:
test: too many arguments
* m4/libtool.m4: Wrap variable in quotes.
If `multilib` is unset (which it apparently is by default), the output
from running a `configure` script contains lines like the following:
```
checking dynamic linker characteristics... ../configure: line 26791: test: yes: unary operator expected
Win32 ld.exe
```
The additional output is not a fatal error. But it could be irritating.
Avoid the unexpected output from `test` by making sure `=` is used as a
binary operator.
* m4/libtool.m4: Prepend "x" on both sides of a comparison with `test`
and the operator `=`.
See also: https://github.com/msys2/MINGW-packages/pull/21905
* m4/ltdl.mk: Add conditional with LTARGZH_EXISTS to append path for
LT_ARGZ_H file to BUILT_SOURCES.
* m4/ltargz.m4: Remove LT_ARGZ_H_PATH.
* m4/ltdl.m4: Add AM_CONDITIONAL for LTARGZH_EXISTS that is set to
true if LT_ARGZ_H is non-empty.
